Th1 Polarization of T Cells Injected into the Cerebrospinal Fluid Induces Brain Immunosurveillance

Approximately 1000–3000 leukocytes per milliliter reside within the human cerebrospinal fluid (CSF) and circulate in the subarachnoid space and ventricular system of the brain and spinal cord. These leukocytes, which are primarily CD4⁺CD45RA⁻CD27⁺CD69⁺ activated central memory T (T_CM) cells (1, 2), migrate from the periphery into the CSF by one of two known mechanisms: 1) crossing the fenestrated endothelium to the choroid plexus stroma and then crossing the choroid plexus epithelium (which is tightly regulated by the blood-CSF barrier) into the CSF; or 2) crossing meningeal vessels before draining into the CSF through the subarachnoid space (2–4).

It has recently been shown in a mouse model of multiple sclerosis (MS) that the initial phase of CNS inflammation is mediated by proinflammatory CCR6⁺ Th17 T cells that are attracted to the choroid plexus upon constitutive expression of the CCR6-ligand CCL20 at this compartment. Consecutively, secondary inflammatory stages in this model involve T cell infiltration via activated venules of the CNS parenchyma (4, 5). Adhesion molecules such as P-selectin, VCAM-1, or ICAM-1 (interacting with P-selectin glycoprotein ligand-1, VLA-4, and LFA-1, respectively, on the T cells) play a key role in mediating the extravasation of the T cells through the blood-brain barrier or the blood-CSF barrier after the T cells are attracted to the CNS (2, 6–8). Such infiltration of T cells into the CSF and perivascular space of the CNS parenchyma may physiologically underlie a routine immune surveillance of the CNS in search for pathogens (9). Accordingly, presentation of Ags at this compartment is presumably required to provide further T cell stimulation before T cells cross the glia limitans or the ventricle ependymal layer into the CNS parenchyma (10–12). This stimulation may ensure that, among the entire lymphocyte repertoire that crosses the blood-brain barrier or the blood-CSF barrier, only those cells that recognize CNS Ags penetrate into the parenchyma.

T cell infiltration into the CNS may occur not only in the context of CNS infection, but also in the context of neurodegenerative diseases. However, although it is believed to underlie the pathogenesis of MS (13–16), in other neurodegenerative processes including brain injury (17, 18), amyotrophic lateral sclerosis (19), stroke (20), and Alzheimer's disease (AD), T cell infiltration into the CNS has been implicated either as exacerbating neurodegeneration (21–23) or, in other instances, as beneficial. Regulatory cytokine profiles of certain T cell subsets (24–26) or the ability of T cells to secrete neurotrophic factors (27, 28) have been suggested as mechanisms underlying the beneficial effects of T cell infiltration into the CNS. However, a key stage of this process, namely the mechanisms responsible for the infiltration of CSF-derived T cells into the CNS parenchyma, is yet unidentified. Elucidation of these mechanisms may greatly assist therapeutic approaches that seek to either block pathogenic entry of T cells into the brain parenchyma or promote a beneficial entry of T cells to specifically target desired Ags within the brain.

In this study, we explored the molecular prerequisites allowing CD4 T cells to migrate from the CSF compartment into the CNS parenchyma following a stereotactic injection of the cells directly into the lateral ventricle. We next characterized the migration pattern, survival dynamics, and function of these cells within the CNS parenchyma in a mouse model of AD exhibiting extracellular deposits of the amyloid-β (Aβ) peptide.

Wild-type C57BL/6 mice were purchased from Harlan. Transgenic (Tg) APP/PS1 (APP-KM670/671NL, PS1-L166P) mice were provided by M. Jucker (Hertie Institute for Clinical Brain Research, University of Tübingen, Tübingen, Germany) (29). APP_Swe/PS1dE9 (APP-K594N/M595L, PS1-exon-9–deleted variant of presenilin 1) Tg mice (stock number 005864), IFN-γR knockout (KO) mice (stock number 003288), OT-II TCR Tg mice (stock number 004194) (30), and Tg mice expressing GFP under the direction of the human ubiqutin C promoter (stock number 004353) were purchased from The Jackson Laboratory (Bar Harbor, ME). All surgical materials, methods, and immunization procedures were reviewed and approved by the Institutional Animal Care and Use Committee of Ben-Gurion University of the Negev (Beer Sheva, Israel; approval number IL-610904).

Mice were killed with an overdose of isofluorane and perfused with cold PBS. Brains were immersed in a 4% paraformaldehyde solution in 4°C overnight, transferred to a 30% sucrose solution in 4°C for 2 d, and fixed in OCT (Tissue-Tek, Torrance, CA). Sagittal sections (35 μm) of the brain were produced with a cryostat and kept in −20°C until used. Sections were rinsed twice in washing solution (0.05% PBS/Tween 20) and permeabilized for 30 min in 0.5% PBS/Triton X-100. Prior to staining, primary Ab diluting buffer (Biomeda, Foster City, CA) was used to block nonspecific binding. Fluorescently stained sections were examined under an Olympus Fluoview FV1000 laser-scanning confocal microscope (Olympus, Hamburg, Germany).

Purified and biotinylated rat anti-CD4 (1:50), rat anti-MHCII (1:50), rat anti-CD54 (ICAM1) (1:50), and hamster anti-CD11c (1:50) were purchased from BioLegend (San Diego, CA). Rabbit anti-Ki67 (1:100) was purchased from Cell Marque (Rocklin, CA). Goat anti-doublecortin (DCX; 1:50) was purchased from Santa Cruz Biotechnology (Heidelberg, Germany). Rabbit anti-mouse Iba-1 (1:1000) was purchased from Wako (Osaka, Japan). Rat anti-mouse glial fibrillary acidic protein (GFAP) (1:250) was purchased from Invitrogen (Grand Island, NY). Rabbit anti-human Aβ Abs (1:500) were generated at our animal facility and examined for specificity by ELISA and immunohistochemistry (IHC). Alexa 488, 546, or 633 Abs (Invitrogen) diluted 1:250–500 were used for secondary staining. TO-PRO-3 (Invitrogen) diluted 1:3000 was used for counterstaining. For the separation of naive CD4 T cells, spleens from naive C57BL/6 mice were harvested and CD4 T cells were separated using the EasySep mouse CD4⁺ T cell enrichment kit (StemCell Technologies, Vancouver, Canada). FACS analysis was performed with the following fluorescent Abs: anti-CD4 (PerCP), anti-CD3 (FITC), anti-CD25 (PE), anti-CD69 (FITC), anti-CD127 (allophycocyanin), anti-CD27 (allophycocyanin), anti-CCR7 (allophycocyanin), anti-CD62L (FITC) (all purchased from BioLegend, San Diego, CA), and anti-CD45RA (PE) (purchased from Miltenyi Biotec, Bergisch Gladbach, Germany). All Abs were used at 0.15 μg per 1 million cells, with the exception of anti-CD45RA (0.1 μg per 1 million cells).

Splenocytes or spleen-derived enriched CD4 T cells were stimulated in 24-well plates (1.5 million cells/well) with 25 μl (cell:bead ratio of 1:1) of mouse T-activator CD3/CD28 Dynabeads (Invitrogen) in complete DMEM (10% FCS, 10 mM HEPES, 1 mM sodium pyruvate, 10 mM nonessential amino acids, 1% Pen/Strep/Nystatin, and 50 μM 2-ME) supplemented with 20 U/ml human rIL-2.

Mice were immunized at 2 mo of age by footpad injection of Aβ_1–42 (100 μg; GenScript, Piscataway, NJ) emulsified in CFA H37Ra (Difco, Detroit, MI). The Aβ_1–42 peptide used for immunization was initially dissolved in a small volume of DMSO to enhance solubility and then diluted to 2 mg/ml in PBS. The peptide was emulsified with CFA to a final concentration of 1 mg/ml. Ten days later, popliteal, inguinal, and iliac lymph nodes were extracted and cells were seeded (5 × 10⁶ cells/ml in 24-well culture dish) in Biotarget medium (Biological Industries) supplemented with 10 μg/ml Aβ_1–42. Every other day thereafter, human rIL-2 (10 U/ml) in complete DMEM was added. Following 1 wk and every 2 wk later, the T cell cultures were restimulated with irradiated (6000 rad) spleenocytes and reseeded (2 × 10⁵ T cells/ml, 5 × 10⁶ irradiated splenocytes/ml in 24-well plates).

Spleens from OT-II OVA TCR Tg mice were cultured with OVA_323–339 (10 μg/ml; GenScript, Piscataway, NJ) and handled similarly to Aβ-specific T cell lines.

For Th cell subpopulation generation, the following cytokines and Abs were added: Th1, anti–IL-4 (20 μg/ml, clone: 11B11; BioLegend) and mouse IL-12 (20 ng/ml; BioLegend); Th2, anti–IFN-γ (20 μg/ml, clone R4-6A2; BioLegend), anti–IL-12 (20 μg/ml, clone C17.8; BioLegend), and mouse IL-4 (4 ng/ml; BioLegend); and Th17, anti–IFN-γ (10 μg/ml), anti–IL-4 (10 μg/ml), mouse IL-6 (20 ng/ml; PeproTech, Rocky Hill, NJ), mouse TGF-β1 (5 ng/ml; PeproTech), and mouse IL-23 (20 ng/ml; R&D Systems, Minneapolis, MN). Recombinant cytokines and cytokine-neutralizing Abs were supplemented in the first three stimulations during seeding and then 2 d later.

T cells (2 × 10⁴) and irradiated (6000 rad) APCs (5 × 10⁵ cells) were cultured in U-shaped 96-well–plate culture dishes in Biotarget medium with the Aβ_1–42 peptide added at increasing concentrations. IL-2 and IL-4 were measured in the supernatant after 24 h, IFN-γ and IL-10 after 48 h, and IL-17A after 72 h, in each case with a sandwich ELISA (BioLegend), according to the manufacturer's instructions. Samples were analyzed with duplicates.

Mice were anesthetized with a sterile mixture of 0.5 ml xylazine, 1 ml ketamine, and 8.5 ml saline, 0.1 ml per 10 g body weight. CD4⁺ T cells were restimulated with 1 μg/ml anti-CD3 for 48 h. Cells were then harvested and resuspended in PBS at a concentration of 50,000 cells/μl. Cells (2.5 × 10⁵) were slowly injected over a period of 5 min into each of the lateral ventricles of the brain with a stereotactic device (coordinates relative to bregma: latero-lateral (x) = +1/−1, dorso-ventral (y) = −0.5, and rostro-caudal (z) = −2.30). Supplemental Fig. 1B and 1C demonstrate the injection site and the brain regions analyzed for the occurrence of CD4 T cells.

Sections were imaged under a confocal microscope, and 1.3-mm² images were imported to Cell Profiler image analysis software (www.cellprofiler.org). Software settings were optimized for identifying only the immunolabeled CD4 T cells, which were then digitally quantified. At least four sections were analyzed per mouse. To appreciate the overall number of infiltrating CD4 T cells in each of the quantified areas (hippocampus, thalamus, and cortex), the average number of cells in a tissue volume of 1 mm³ is shown in Supplemental Table I.

Brain sections were immune labeled with anti-CD4 and anti-Ki67 Abs to view under a confocal microscope. The total number of CD4 T cells within each section was measured with the Cell Profiler image software, whereas the number of CD4⁺Ki67⁺ cells within the same image was counted manually. The percentage of CD4⁺Ki67⁺ cells of the total CD4⁺ cells was then calculated.

Brain sections were stained for apoptotic cells with the TUNEL (MEBSTAIN Apoptotic Kit Direct; MBL, Nagoya, Japan) and counterstained with TO-PRO-3 for detection of cell nuclei. As a positive control for the kit, brain sections from a mouse model of stroke were similarly stained and imaged. The dentate gyrus/hippocampus area of at least four sections from each brain was imaged with a confocal microscope. All TUNEL⁺TOPRO-3⁺ cells were counted manually.

Naive CD4 T cells were collected using a CD4 magnetic separation kit (StemCell Technologies, Vancouver, Canada) with typical results of CD4⁺CD25⁻CD69⁻ cells in the range of 75% of separated cells. These cells were injected into the lateral ventricle of wild-type (WT) mice or were activated for 2 d in vitro using mouse T-activator CD3/CD28 Dynabeads (Invitrogen, Grand Island, NY) and injected thereafter. Mice were killed at 5 d postinjection (dpi), and their brains were processed for IHC. At least four sections from each mouse were stained for CD4 T cells, and their numbers were calculated using the Cell Profiler image software.

Quantification analysis of Aβ plaques in the brain was performed in two sections (35 μm thick) per hemisphere stained for Aβ. In each section, three regions from the cortex and one from the hippocampus were taken for quantification. Fluorescence intensity was first obtained in sections from control mice (immunized with adjuvant only), and identical laser-scanning parameters were then used for the entire experiment. Using Volocity 3D image analysis software (Improvision, Waltham, MA), an intensity threshold was set to mark only those areas showing significant staining. The average fluorescent area per brain section was calculated for each of the analyzed groups.

Quantification analysis of Iba-1 and MHCII in the brain was performed in two sections (35 μm thick) per hemisphere immuno labeled for Iba-1, MHCII and ICAM-1. Two regions from the lateral ventricle wall of each section and, for ICAM-1 only, two regions from the choroid plexus were imaged for quantification with a confocal microscope. Fluorescence intensity was first obtained in sections from control mice (intracerebroventricular [ICV] injected with PBS), and identical laser-scanning parameters were then used for the entire experiment. Using the Volocity 3D image analysis software (Improvision), an intensity threshold was set to mark only those areas showing significant staining. The sum of fluorescent intensity was calculated for each image, and an average intensity was calculated for each mouse. All results were normalized compared with the control treatment.

Neurogenesis was evaluated by counting DCX⁺ cells in the dentate gyrus in 35-μm brain sections across z-stack images taken at 1-μm intervals. Early-stage progenitors (subgranular DCX⁺ cells) and total DCX⁺ cells in the dentate gyrus were counted in three to four sagittal sections evenly distributed in the analyzed brains. Early-stage progenitor cells were distinguished by their subgranular location and the horizontal position compared with vertical position with longer processes in the granular layer.

The pattern of Aβ T cell distribution was calculated using the L function method, as previously described (31). Briefly, confocal TIFF images of brain sections immunolabeled for CD4 T cells were imported to the Cell Profiler software to measure the T cell coordinates. The R statistical package for spatial statistics (SPATSTAT; http://www.spatstat.org/spatstat/) was then used to calculate the L function, a variance-stabilized version of the Ripley’s K function, which allows detecting a deviation from a homogeneous spatial point distribution. The K function equation is as follows:

where d_ij is the distance between the i^th and j^th points in a data set of n points; λ is the average density of points generally estimated as n/A, where A is the area of the region containing the examined points. We selected all the CD4 T cells in each brain image of the hippocampus, thalamus, and cortex (at least four images from each region per mouse) from APP/PS1 Tg (n = 3) and WT mice (n = 4) to calculate the L function as follows:

The L function and confidence envelope (generated by Monte Carlo simulation) were first calculated for the distribution of Aβ T cells in the brain of WT mice, which served as a reference for the distribution of Aβ T cells in the brain of APP/PS1 Tg mice.

Mice were perfused with PBS, and half brains were immediately frozen in liquid nitrogen and stored in −80°C. RNA was extracted by phenol-chloroform procedure and analyzed by Bioanalyzer (Molecular Research Center). A total of 2 μg RNA was reverse transcribed with high capacity cDNA reverse transcription kit (Applied Biosystems, Invitrogen), and 125 ng cDNA was used for quantitative real-time PCR analysis. Except for SIRP-1β and CXCL9, all genes were analyzed with TaqMan Gene Expression Assay (Applied Biosystems, Invitrogen). The Cxxc1 gene was used as endogenous control to normalize gene expression. SIRP-1β and CXCL9 expression was analyzed with Power SYBR Green PCR mix (Applied Biosystems, Invitrogen) with β-tubulin as endogenous control. The primers used were as follows: SIRP-1β, forward, 5′-CCCGTTCACAGGAGAACATT-3′, and reverse, 5′-CCGGAGACCATAGGTGAAGA-3′; CXCL9, forward, 5′-CCTGGAGCAGTGTGGAGTTC-3′, and reverse, 5′-AGGCAGGTTTGATCTCCGTTC-3′; and β₂ microglobulin, forward, 5′-TGGTGCTTGTCTCACTGACC-3′, and reverse, 5′-TATGTTCGGCTTCCCATTCT-3′.

All statistical analyses were performed with GraphPad Prism version 5.02 for Windows (GraphPad Software, San Diego, CA). All variables are expressed as means ± SEM or SD, as indicated in figure legends. The p values were calculated with Student t test or Mann–Whitney U test, as indicated in figure legends.

The process of T cell entry into the CNS requires preactivation of the cells in peripheral lymph nodes, whereby the cells upregulate adhesion molecules and tissue-specific chemokine receptors (4). We first explored whether T cell preactivation is essential for the migration of CD4 T cells from the lateral ventricles into and within the CNS parenchyma. To this end, we ICV injected naive or preactivated (following polyclonal anti-CD3/anti-CD28 stimulation in vitro) spleen-derived CD4 T cells stereotactically into the lateral ventricles of wild-type (WT) autologous C57BL/6 mice (2.5 × 10⁵ cells/hemisphere) (see 2Materials and Methods and Supplemental Fig. 1B, 1C). Brains were processed for IHC analysis of CD4 T cells in different brain regions 5 dpi.

On average, only a small number of T cells is found in the hippocampus, thalamus, and cortex of mice ICV injected with naive T cells (13.3 ± 3.2, 2.5 ± 1.3, and 7 ± 1.7 cells per section, respectively). A polyclonal in vitro preactivation of the cells, however, significantly increases their migration and accumulation in the hippocampus and cortex (67.5 ± 20.4 and 113 ± 34.83 cells per section, respectively), and to a lesser extent (30.9 ± 8.5 cells) in the thalamus (Fig. 1A, 1B, Supplemental Table I). ICV injection of polyclonal-activated GFP⁺ CD4 T cells shows that most CD4 T cells infiltrating the brain parenchyma at 5 dpi are not endogenous T cells (i.e., recruited via the choroid plexus or the brain vasculature) but are the GFP⁺ injected cells (Supplemental Fig. 1D, 1E).

To determine the role of chemokine signaling in CD4 T cell migration, we stereotactically injected preactivated OVA-specific CD4 T cells into the lateral ventricles of WT mice. This was done with or without pretreating the cells in vitro with pertussis toxin (OVA + pertussis toxin [PTX] and OVA − PTX cells, respectively), which irreversibly blocks Gα_i signaling and leads to chemotaxis arrest (32). Whereas the OVA − PTX cells readily migrate into the brain parenchyma, OVA + PTX cells are observed only within the ventricle borders and leptomeninges (Fig. 1C), although the PTX treatment does reduce T cells’ cytokine secretion in vitro (Fig. 1D). Chemokine signaling appears to mediate T cell migration not only to, but also within, the brain parenchyma, as a stereotactic injection of OVA − PTX, but not OVA + PTX, cells directly into the cortex results in T cell migration within the parenchyma (Fig. 1E). Indeed, viable cells with a polarized cell body (typical of a migratory T cell phenotype) are found close to the injection site of OVA − PTX cells; however, only dead cell particles, possibly phagocytosed by microglia, are found close to the injection site of OVA + PTX cells.

We next tested whether ICV-injected pro- and anti-inflammatory T cell subsets have similar migratory capabilities. We therefore generated Aβ-specific CD4 T cells with a primarily effector memory phenotype and differentiated them into Th1, Th2, or Th17 T cells (Supplemental Fig. 2A–C). Cells from each line were then ICV injected into 2-mo-old WT mice, which were killed at 5 or 28 dpi for IHC and quantitative PCR (qPCR) analysis of the brain tissue. Analysis of CD4 T cells within the hippocampus, thalamus, and cortex reveals that migration into the brain parenchyma is significantly greater for Th1 cells than for Th2 or Th17 cells, and that the number of all T cell subsets is markedly decreased from 5 to 28 dpi (Fig. 2A–C). The estimated number of CD4 T cells within a tissue volume of 1 mm³ in the quantified regions suggests that at least 20% of the injected cells infiltrated the brain (Supplemental Table I). Accordingly, a qPCR analysis at 5 dpi shows a significant upregulation of the mRNA of Th1 cytokines (namely IFN-γ and TNF-α) in mice ICV injected with Th1 cells (Fig. 2D). Except for IL-10 mRNA, which is slightly induced in the brain of mice ICV injected with Th1 or Th2 cells (Fig. 2D), the Th2 (IL-4) or Th17 (IL-17) mRNAs are not detected in significant amounts in the brain of any of the mouse groups. Chemokine expression analysis reveals that, except for CCL5, which is induced by ICV injection of Th1 as well as Th2 cells, and CCL2, which is induced by ICV injection of all three T cell subsets, the chemokines CXCL9, CXCL10, and CXCL11 (the last three are ligands of the CXCR3 receptor on Th1 T cells) are induced only in mice ICV injected with Th1 T cells (Fig. 2E, Table I). The Th2 chemokine CCL11 is induced only in mice ICV injected with Th2 T cells (Fig. 2E, Table I), and the chemokines CCL3, CCL4, CCL20, and CXCL12 are not induced by either Th1, Th2, or Th17 T cells (Fig. 2E, Table I). An ICV injection of Th1 T cells to IFN-γR KO mice results in neither cytokine nor chemokine mRNA induction (Fig. 2D, 2E), suggesting the following: 1) IFN-γ plays a key role in inducing the cytokines and chemokines; 2) both Th1 and Th2 chemokines are induced by IFN-γ; and 3) T cell function and migration within the brain parenchyma depend on IFN-γ signaling to the neural tissue.

We next characterized the infiltration process at the lateral ventricle wall facing the hippocampus directly across the ependymal layer. Fig. 3A shows, in the brain of a mouse ICV injected with PBS, GFAP⁺ astrocytes forming subependymal elongated processes, Iba1⁺ microglia extending processes protruding to the ventricle ependymal layer, and ICAM-1 expressed primarily by the ependymal cells facing the ventricle lumen. Following an ICV injection of Th1 cells, ICAM-1 levels within the luminal surface of the ventricle wall are significantly increased (Fig. 3A, Supplemental Fig. 3A, 3B) and both the ependymal and microglial cells at the ventricle wall express MHCII and interact with the infiltrating T cells (Fig. 3B). In addition, compared with mice ICV injected with Th2 or Th17 T cells, mice ICV injected with Th1 T cells show significantly increased levels of Iba-1 and MHCII that are coexpressed by microglia/monocytes integrated with the ependymal layer from both the luminal and parenchymal sites of the ventricle wall (Fig. 3B–D). Notably, ICAM-1 and MHCII are also upregulated within the choroid plexus colocalized with CD4 T cells (Fig. 3E) in mice ICV injected with all Aβ-reactive T cell subsets (namely Th1, Th2, or Th17 T cells) (Supplemental Fig. 3C). However, primarily Th1 and to a lesser extent Th2 and Th17 T cells induce MHCII expression by APCs within the choroid plexus (Fig. 3E, Supplemental Fig. 3C). A qPCR analysis reveals that CD74 (the Li invariant chain) and to a lesser extent ICAM-1 are significantly induced in mice ICV injected with Th1 cells but not in mice ICV injected with Th2 or Th17 T cells (Fig. 3F). An ICV injection of Th1 T cells to IFN-γR KO mice does not increase CD74 and ICAM-1 mRNAs (Fig. 3F), suggesting that this induction is due to the increased expression of IFN-γ. Taken together, these data demonstrate the key role of IFN-γ in inducing ICAM-1 and MHCII at the ventricle, which, in concert with chemokine expression, facilitate the infiltration of T cells in the CSF, and most prominently of Th1 cells, into the brain parenchyma and the choroid plexus.

Because primarily Th1 T cells transmigrate into the brain parenchyma, we next asked whether Aβ-specific Th1 cells are able to specifically target the Aβ Ag deposited extracellularly as plaques in the brain of an AD mouse model. We thus ICV injected Aβ-specific Th1 T cells to 5-mo-old WT mice (Aβ→WT mice) or to APP/PS1 transgenic mice as a model of AD (Aβ→AD mice) and analyzed the brains for CD4 T cells and Aβ plaques at 28 dpi. In addition, to determine the role of Ag specificity of the injected Th1 cells, we ICV injected OVA-specific T cells to APP/PS1 Tg mice (OVA→AD mice).

Significantly more CD4 T cells are observed at 28 dpi in the hippocampus, thalamus, and cortex of Aβ→AD mice as compared with Aβ→WT mice and with OVA→AD mice (Fig. 4A, Supplemental Table I). An L function analysis (see 2Materials and Methods) shows that for interpoint distances >10 μm, the L function deflects upward in Aβ→AD mice compared with Aβ→WT mice in all three brain regions (84, 63, and 89% of the images in the hippocampus, thalamus, and cortex, respectively; Fig. 4B), indicating that the T cells are more clustered in Aβ→AD mice. A detailed IHC analysis of the hippocampus reveals that, whereas the cells are randomly distributed in Aβ→WT mice (Fig. 4C), they are clustered around Aβ plaques in Aβ→AD mice at 28 dpi (Fig. 4C, 4D). In addition, in Aβ→AD compared with OVA→AD mice, MHCII is increased and is colocalized with T cells at the sites of Aβ plaques (Fig. 4E). This targeting of MHCII⁺ cells at plaque areas does not, however, significantly enhance T cell proliferation because immunolabeling of the brains of Aβ→WT and Aβ→AD mice with the Ki67 proliferation marker at 5, 14, and 28 dpi reveals similar proportions of CD4⁺Ki67⁺ T cells (Fig. 4F, 4G). Taken together, these results suggest that the accumulation of Aβ in the brain of APP/PS1 Tg mice promotes the targeting of T cells specifically to their Aβ Ags and thereby increases MHCII expression, which presumably facilitates the longer retention of these cells in the brain.

In light of these results, we next sought to determine whether ICV-injected T cells can facilitate plaque clearance as part of their effector function in the brain. We therefore ICV injected Aβ-specific Th1 cells or PBS to the APP/PS1-dE9 transgenic mouse model of AD (Aβ→AD and PBS→AD, respectively) aged 12–15 mo. Brain sections were immunolabeled with anti-Aβ at 28 dpi, and the areas occupied by Aβ plaques in the hippocampus and cortex were analyzed. Compared with PBS→AD mice, a 56 and 30% reduction in plaque load is observed in the hippocampus and cortex, respectively, of Aβ→AD mice (Fig. 5A, 5B). A RT-PCR analysis indicates that MHCII, IFN-γ, and TNF-α in the brains of Aβ→AD mice remain significantly upregulated as compared with PBS→AD mice at 28 dpi (∼3-fold), however, to a markedly lower extent than at 5 dpi (Fig. 5C, Table II). Similarly, SIRP-1β, which was shown to increase Aβ uptake by microglia, is induced by Th1 cells (but not by Th2 or Th17 cells) at 5 dpi (Fig. 5D) and to a lesser extent at 28 dpi (Fig. 5C). In addition, of all the chemokines induced at 5 dpi, only CCL5 and CXCL9 remain upregulated. These data suggest a very low-grade inflammatory reaction induced by Aβ plaque–associated Th1 T cells sufficient to enhance the clearance of Aβ by microglia.

We used two different approaches to determine the effects of ICV-injected Aβ-specific Th1 T cells on neuronal viability. First, we quantified the amount of apoptotic cells in APP/PS1-dE9 Tg mice ICV injected with PBS or with Aβ-reactive Th1 T cells (PBS→AD and Aβ→AD mice, respectively) using the TUNEL staining. Sections taken from a mouse brain following vascular injury were used as a positive control for TUNEL staining. The analysis shows that the number of apoptotic cells is very low and similar in the brain of Aβ→AD mice and of PBS→AD mice, whereas the positive control shows markedly elevated apoptosis (Fig. 6A, 6B). Second, the amount of proliferating and differentiating neuronal progenitors in the hippocampus was measured in brain sections taken from 6-mo-old adult WT mice and from 10- to 12-mo-old APP/PS1-dE9 Tg mice, each left untreated or ICV injected with Aβ-specific Th1 T cells or with PBS. Sections were immunolabeled with anti-DCX, a microtubule-associated protein expressed by neuronal precursor cells. The total number of DCX⁺ cells is similar in all groups (Fig. 6D). However, whereas the number of subgranular neuronal progenitors (see arrows in Fig. 6C) is also similar in all WT groups (Fig. 6E, left panel), it is significantly increased in 10- to 12-mo-old APP/PS1-dE9 Tg mice ICV injected with Th1 T cells (Fig. 6E, right panel). Overall, these data demonstrate that ICV-injected Th1 T cells induce neither neuronal loss nor chronic neuroinflammation, and that in APP/PS1-dE9 Tg mice they induce a temporary increase in dentate gyrus proliferating neuronal progenitors (Fig. 6).

We hereby provide evidence that, in mice, activated CD4 T cells polarized toward the Th1 phenotype and ICV injected into the lateral ventricles effectively survey the brain parenchyma. The migration of these cells within the neural tissue is chemokine receptor dependent and is associated with ectopic increased expression of ICAM-1 and MHCII. We also report that ICV-injected Aβ-reactive Th1 cells effectively target Aβ plaques in a mouse model of AD, and that the migrating cells are retained in significant numbers in plaque-specific regions for ∼4 wk. The Th1 cells are active within the brain throughout this period in a manner that supports their migration pattern and function, and are able to induce the clearance of Aβ plaques. This process is accompanied by enhanced neurogenesis with no apparent chronic brain inflammation or neuronal loss.

It was recently demonstrated in mouse models of MS and AD that, following immunization or adoptive transfer of T cells, Ag presentation by dendritic cells at the perivascular space of meningial and parenchymal vessels is a prerequisite for CD4 T cell transmigration into the CNS parenchyma (10, 11, 33–35). Whether such Ag presentation is required at the lateral ventricle before CD4 T cells can migrate into the brain parenchyma was, however, not clear. In this study, we demonstrate that, following in vitro activation and ICV injection, both OVA-specific and polyclonal CD4 T cells migrate into and within the brain parenchyma. This suggests that activated T cells do not require further Ag presentation prior to crossing the ependymal layer into the brain parenchyma. Because the majority of T cells in the CSF exhibit a central memory phenotype (T_CM), a two-step process can be considered (Fig. 7), as follows: 1) Effector and/or effector memory T (T_EM) cells migrating into the CSF compartment during peripheral inflammation can readily cross the epithelial layer and/or the subarachnoid space into the CNS parenchyma without further Ag presentation. 2) In the case of Ag recognition, the infiltrating T cells locally stimulate APCs (such as CD11c⁺ cells) that are located at the subependymal area and have processes protruding into the ependymal layer of the ventricles (36). This stimulation, in turn, promotes the activation of Ag-specific CSF T cells to allow their infiltration from the ventricle into the brain parenchyma. Whereas in the first step Ag specificity is determined in peripheral lymph nodes according to the pathogenic trigger, in the second step Ag specificity is determined in the CNS according to Ag drainage from the CNS parenchyma. The mechanisms whereby CSF T_CM cells exert effector functions in the CNS need, however, further research.

What is the molecular machinery facilitating effective migration of the activated T cells from the ventricles into the brain parenchyma? A recent study by Reboldi et al. (5) demonstrates that Th17 T cells expressing CCR6 pave the way for T cell infiltration into the parenchyma. The authors reported that the CCR6-ligand CCL20 is constitutively expressed by epithelial cells in the choroid plexus and suggest that the migration of Th17 T cells into this compartment is required for the initiation of experimental autoimmune encephalomyelitis. However, beyond the initial phase and upon disease progression, primarily CCR6⁻ cells are found in the brain, suggesting that effector T cells other than Th17 may enter the parenchyma. In light of this study, we demonstrate in this work that, whereas significant numbers of ICV-injected Th1 T cells cross the epithelial barrier of the ventricles, only a few ICV-injected Th2 and Th17 T cells are found in the brain parenchyma. Although not conclusive, it is likely that the predominant expression of IFN-γ by Th1 T cells underlies their migratory capability within the tissue (37, 38). Indeed, IFN-γ mRNA is abundantly expressed in brains ICV injected with Th1 T cells and is associated with upregulation of MHCII as well as of key migration-promoting cytokines and chemokines. In addition, both ICAM-1 and MHCII are upregulated in the tissue upon the ICV injection of Th1 cells and are colocalized with the T cells at both the ventricle epithelium and within the parenchyma. Taken together, we suggest that, whereas Th17 T cells preferentially infiltrate the ventricles and subarachnoid space [e.g., following infection or experimental autoimmune encephalomyelitis induction (5)], Th1 cells are more effective in infiltrating into the parenchyma to target their Ag, which is in line with previous findings showing IFN-γ–expressing CD4 T cells (within both T_EM and T_CM cell subsets) in the CSF of patients with neuroinflammatory diseases (39).

Because neither an upregulation of IFN-γ–dependent genes nor IFN-γ itself is detected in IFN-γR KO mice ICV injected with Th1 T cells, we suggest that the entire gene expression induced (e.g., chemokines, cytokines, CD74, and ICAM-1) results from effective T cell migration rather than from their prior in vitro stimulation. Thus, once the Th1 T cells initiate the migration process within the brain parenchyma, they receive signals that maintain the expression of IFN-γ and thereby its gene targets. Our data thus support previous studies showing that the migration process is MHCII independent and is, rather, adhesion and chemokine dependent (40). The Th1 T cells therefore pave the way by inducing adhesion molecules and chemokines that allow them effective migration within the tissue in their search for Ags. This process presumably also allows the recruitment of other T cell subsets and monocytes that participate in the immune response and its resolution.

We also demonstrate that Aβ-reactive Th1 cells effectively target Aβ plaques in the brains of APP-Tg mice but are randomly distributed, and mostly do not persist longer than 28 dpi in the brains of WT mice. OVA-specific T cells, in contrast, fail to cluster around Aβ plaques in the AD mouse model, suggesting that Aβ was presented to the T cells via colocalized MHCII^high APCs that were either differentiated from brain-endogenous microglia or recruited from the blood due to the increased expression of CCL2 (41, 42). These observations suggest that, following the migration of T cells within the brain parenchyma, the cells either undergo apoptosis, drain from the tissue, or retain at inflammatory foci, where their cognate Ag is presented by APCs. Such retention of IFN-γ–secreting T cells at the plaque areas locally impacts on gene expression, which promotes further recruitment of T cells to the plaques and enhances the phagocytic activity of microglia, with, however, only a very mild expression of proinflammatory cytokines such as TNF-α and IL-1β. Importantly, IFN-γ–secreting CD4 T cells not only did not induce the expression of IL-17 in the brain, they induced upregulation of beneficial factors such as IL-10 as well as chemokines, which perhaps pave the way for regulatory T cells. Recent studies indeed highlight the key immunoregulatory functions of IFN-γ (43–45) as opposed to the immunopathogenic functions of IL-17 (46).

Previous investigations have shown that a single therapeutic Aβ immunization of APP/IFN-γ double-Tg mice, which express limited amounts of IFN-γ in the brain, results in T cell migration to Aβ plaques associated with significant plaque clearance (47). However, similar to human AD patients immunized with Aβ (Elan’s trial), the immunized mice develop meningoencephalitis presumably due to T cell activation at the perivascular/subarachnoid space prior to crossing the glia limitans. Similar effects of T cell activation at the brain vasculature were also described in MS (14) and are likely to impair the brain-endogenous capacity for cell renewal and repair (48, 49). In contrast, because activated T cells in the current study were injected directly into the lateral ventricle, no further activation was required prior to crossing the epithelial layer or the vasculature’s glia limitans into the brain parenchyma, and, as such, the injected cells induced neither significant vascular pathology nor cellular apoptosis, but instead, similar to previous studies demonstrating the positive effects of T cells and IFN-γ on neurogenesis (24, 27, 50, 51), slightly enhanced hippocampal neurogenesis in an age-dependent manner. Given recent observations pointing to age-related Th2 shift in the circulation (52) and at the brain compartment (53), it is possible that brain immunosurveillance and neural repair are impaired with aging and further in age-related neurodegenerative diseases. The migration capacity of IFN-γ–expressing T cells secreting a variety of beneficial T cell proteins (such as cytokines, chemokines, and neurotrophic factors) that cannot be delivered into the brain at sufficient concentrations thus urges for further studies examining their safe therapeutic potential within the aged CNS milieu.

We thank Dr. Ram Gal for valuable comments throughout the manuscript.

A.M. has a pending patent application.